Spectral response is an important parameter of satellite instruments, and is the integrated performance of the reflective and transmission characteristics of each optical element and the spectral response characteristics of detectors. As do not have monochromatic light source devices for onboard spectral calibration,the spectral response in orbit is usually characterized by prelaunch evaluation in laboratory. Due to lack of vacuum environment, spectral response measurement in laboratory will be certainly contaminated due to atmospheric absorption between the calibration source and the instrument. In order to improve spectral calibration accuracy, influences of atmosphere contamination on the spectral response measurements of FY-2 water vapor channels were assessed based on a series of sensitivity tests using radiative transfer models (RTM). Because of the atmospheric absorption, the spectral responses become unsmooth, and the structure shows good consistency with the transmittance distribution. The channel brightness temperatures (BT) simulated based on contaminated SRF will be overestimated, and the biases increase exponentially over water vapor column. The biases are greater than 0.5 K and less than 1 K only during extremely dry environment with humidity less than 15% and homogeneous path less than 1m. However, when the homogeneous path approaches 4m under a normal ambient condition where the humidity is around 35%, the simulation BT bias can reach 4 K. Atmospheric contamination on spectral response measurement is happen for all water channels of FY-2 series satellites. Spectral correction was implemented for FY-2D by removing horizontal transmittance influence using spectra ratio. The corrected spectral response curve becomes much smooth and the distribution is more physically reasonable. The contamination leads to a warm bias of 2.2 K for clear sky top of atmosphere BT simulation, and a 7.6% positive bias for the blackbody radiance as well as the radiometric calibration calculation. The influence of atmospheric contamination on spectral response measurement is serious, thus can’t be ignored not only for the water vapor channel but for all absorptive channels. The contamination to spectral measurements which can not be neglected should be corrected from dividing the original spectral response by atmospheric transmittance.